The R-COOL study, funded by Echord++ EU program, is a preliminary study in chronic low back pain patients. Its main objective was to evaluate the feasibility of a humanoid robot supervision of a set of movement performed in a rehabilitation program.
This prospective, randomized, controlled, single-blind and bi-centric study compared a usual rehabiliation program to the same one from which 30 minutes a day of physical activity were substituted by robot supervised movements. The primary endpoint was the time spent for physical activity. The secondary endpoints were lumbar pain, disability, fear and beliefs related to physical activity, acceptability and adverse events. The control group practiced 3 hours per day 5 days a week during 3 weeks a conventional rehabilitation program. The robot group performed 2.5 hours of the same treatment and 30 minutes of rehabilitation supervised by the robot.
30 patients were randomized, 15 in each group. 3 patients were excluded from the analysis in the robot group. The duration of rehabilitation did not differ between groups. No statistically significant differences between groups were observed for the secondary endpoints except for patient satisfaction which was lower in the robot group. No severe adverse event was reported.
Robot supervised exercises can be inserted in rehabilitation programs in chronic low back pain patients. Although technology driven physical activity programs have not demonstrated to improve the medical condition of these patients more than conventional therapy, they may nevertheless reduce costs and enhance the number of patients beneficiary of these programs.
Objectives : Chronic low back pain (CLBP) is one of the leading causes of disability and a major public health problem because of its prevalence and its socio-economic impact. It is now recognized in the literature, with strong evidences, that physical activity alleviates pain and disability and therefore improves the quality of life. However, the patient’s adherence to the physical activity program realization decreases over time, which is likely to lead to lower efficiency and relapse. It is therefore necessary to improve patient involvement in the rehabilitation program and to promote physical activity. Meanwhile, in recent years, there has been an explosion of electronic devices for coaching physical activities, both in the general public and in rehabilitation. Several studies show that electronic devices are promising to promote involvement and learning in terms of health, physical activity or social behavior. But few studies have been published to assess their efficacy and acceptability. The R-COOL study is a preliminary study which main objective was to evaluate the feasibility of a humanoid robot's supervision of rehabilitation exercises series for chronic low back pain patients.
Methods : This prospective, centrally randomized, controlled, single-blind and bi-centric study was conducted from October 2017 to May 2019 in two rehabilitation centers of Brittany (France). It is a part of the KERAAL project (funded by the ECHORD program of European community : European Clearing House for Open Robotics Development). The primary endpoint was the average daily time of physiotherapy. A 30mn session during which a set of stretching movements was coached by the robot without a physiotherapist replaced a 30-minute period of the conventional rehabilitation program. This program was adapted to the patient’s needs but contained at least 180mn of daily physiotherapy, physical therapy and occupational therapy.
The Robot Supervised Rehabilitation Group (RG) performed each day, 5 days a week, during 3 weeks, 2.5 hours of conventional rehabilitation and 30 minutes of rehabilitation supervised by the robot called POPPY. The control group (CG) performed a usual rehabilitation program with physiotherapy, occupational therapy, adapted physical activity and balneotherapy with the same schedule with a supplementary 30 mn physical activity session instead of robot supervised session.
Secondary outcomes were lumbar pain evaluated by the Visual Analogue Scale (VAS), disability evaluated by the Roland-Morris Questionnaire (RMQ) and the Dallas Questionnaire (DQ), fear and beliefs related to physical activity evaluated by the Fear Avoidance Beliefs Questionnaire - French version (FABQ fr), acceptability assessed by a questionnaire for patients and for physiotherapists and adverse events. These outcomes were evaluated at Day 0 (D0), Day 21 (D21) and Month 6 (M6). The first acceptability questionnaire was administrated at Day 2 (D2), in order to have an experience of the robot).
POPPY is an open source humanoid robot with a spine that includes 5 motorized degrees of freedom allowing the realization of movements used in low back pain rehabilitation program. POPPY is also able to interact with its environment by a voice synthesizer and a display screen included in the head of the robot mimicking facial expressions. It’s connected to a Kinect camera providing in real time the 3D position and orientation of the patient’s body segments. Stretching movements proposed during the rehabilitation program was learned to the robot. For this purpose a specific software was developed as part of the KERAAL project.
Results : Among the 31 patients included, 30 were randomized either in the RG or in the CG , with 15 patients per group. In the RG, 3 patients were excluded from the analysis, one due to refusal to consent and 2 because they withdrew their consent during the study and for whom we only have baseline data. There was no significant difference between groups at baseline.
No statistically significant difference of the average daily time of physiotherapy was observed during the study period between groups (RG : 228.8 minutes [190.7-266.9]; CG :226.5 minutes [190-262.9], p=0,92). No statistically significant differences between groups were observed for the secondary outcomes. 20 subjects out of 27 were evaluated at 6 months. We did not compare post-treatment and M6 assessments because too many confounding factors exist in these chronic pain subjects. No between group difference was observed except for the anxiety/depression sub-score of the DQ that was statistically lower in the control group at M6. All the parameters were in the range observed between D0 and D21. No serious adverse event was reported.
The estimation of the acceptability of the robotic system was divided in two categories: questions about utility and ease of use. The robot usability was estimated strongly lower than that of the conventional therapy. Questions focused on the ease of use had better results.
Conclusion : In agreement with our hypothesis, the duration of rehabilitation sessions assisted by the robot is equivalent to that of the sessions performed with the therapist. At this stage, evaluation of the device by patients and physiotherapists reports an ease of use of the device but utility has not been demonstrated. However, our study is conducted as part of a technological research project with a limited level of completion of the robot, that could cause patient dissatisfaction. The type of sensor for appropriate feedback needs further development and an accurate way of human robot interaction to be implemented.
olivier Remy-Neris– Professor, CHRU Brest
Agathe Blanchard– medical intern, CHRU Brest - Hopital Morvan
Mai Nguyen– Lecturer, IMT Atlantique
Mathieu Simonnet– Engineer, IMT Atlantique
Xavier Mallauran– Doctor, Fondation Ildys
Myriam Le Goff-Pronost– Lecturer, IMT Atlantique